Contents:

Free training: "Practical course on neural networks"
Find out moreArtificial neural networks are actively creating a variety of text materials, including articles, social media posts, essays, official documents, announcements, and even personal messages. Their influence is felt everywhere. In this context, the question arises: can we confidently distinguish between texts created by machines and human works? We tested three services designed to identify generated texts and also asked neural networks to distinguish between texts written by people and robots. As a result, we compiled a list of characteristic features that will help you visually recognize when a text is the product of a neural network.
Contents
- Tools for analyzing text generated by artificial intelligence.
- How do neural networks determine whether text is written by a person or created by a machine?
- There are several methods for checking text for errors and compliance without resorting to online services. One of the simplest ways is careful reading. Sometimes it is worth reading the text aloud to catch minor flaws and unnatural wording. It is also useful to give the text to someone else for evaluation; a fresh look can reveal errors that you might have missed.
Another effective method is to use an editor. For example, you can use the built-in spell-checking features in a word processor such as Microsoft Word or Google Docs. These programs usually have tools for identifying spelling and grammar errors.
Also, it's worth paying attention to the structure of the text. Check the logical structure of the paragraphs, the sequence of thoughts, and the presence of connecting words. It is important that the text is understandable and readable.
Don't forget about grammar rules and stylistic norms. Knowing the rules of punctuation and syntax will help you avoid common mistakes. If in doubt, you can refer to reference books or language textbooks.
Thus, careful editing, using the capabilities of text editors, and referring to classic reference materials will help you thoroughly check the text without the need for online services.
Tools for analyzing text generated by artificial intelligence
There are many platforms designed to analyze texts for their creation using neural networks. However, it's worth noting that a text generated based on a carefully crafted query can be quite similar to a human-written work. If the author is familiar with the workings of generative artificial intelligence, they can formulate instructions that will imbue the text with features characteristic of their individual style. For example, he can provide role models, set rules for punctuation, syntax, language complexity, and other important aspects.
Perhaps the author should not be condemned for using neural networks if he has invested significant effort in creating a text that is difficult to distinguish from a human work.

Read Also:
Efficient queries for neural networks: recommendations for formulating questions for ChatGPT and other AI models
To test the services, we will test two texts dedicated to the history of Bitcoin: one written by the author, and the other created using Gemini. Each text will be uploaded to the service in turn to evaluate their quality. We will consider the test successful if the service identifies the author's text as a human work, and the generated one as the result of the neural network.
If you want to reproduce our experience, here are the materials: one created by a person, the other using a generator.
The service from Sber is focused exclusively on texts in Russian and successfully accomplishes this task. When the project launched in the summer of 2024, the creators guaranteed an accuracy level of up to 94.7%.
Currently, GigaCheck can be used free of charge both on the official website and through the Telegram bot. Developers provide responses to integration requests in accordance with submitted applications. The maximum text size for analysis is 10,000 characters.


There is another free service worth testing. Copyleaks supports 30 different languages and guarantees an impressive accuracy of over 99%. It allows you to check texts up to 25,000 characters long. This service also has an extension for Google Docs, which allows you to check directly in the document. However, this service is paid for regular use.


PR-CY is another Russian online service that receives positive reviews online. This is a paid resource, and, unfortunately, full testing with large volumes of data cannot be carried out without payment. To simplify testing of two texts, we decided to shorten them a little.


AI Detector Writer is a free online tool for detecting artificial intelligence. The developers have provided users with generous capabilities: the service can process texts up to 350,000 characters long and allows uploading text files.


GPTZero is one of the first tools for detecting artificial intelligence. This service offers support for various languages, including Russian. If the text exceeds the input field limits, users can upload a text file of the required size. In addition, this service has an extension for Google Docs.
Let's conduct a practical test of the detector using the same texts.


Our study is not an in-depth analysis - we tested only two texts using detectors. We did not consider materials partially generated by artificial intelligence or neural network texts that had been edited by humans. However, the results of this test suggest that relying on services to detect the presence of AI in texts should be done with caution.

Read also:
Seven neural networks for developers: ways to speed up and improve the coding process.
If you are interested in reproducing our experience, Here are the texts: one is original, the other is generated.
Methods by which neural networks recognize the differences between texts created by people and machines
Let's ask neural networks to determine the author of the text.
The first text, created by a person, was characterized by ChatGPT as a neural network text.
This text was created using a neural network, and here are the reasons for this.
The text has a logical sequence and flow, but its uniformity raises some doubts. Phrases are of similar length, the sentence structure practically does not vary, and the transitions between paragraphs are created with such care that they seem mechanical. A writer with a developed style has intonation changes, small pauses in the semantic context, and variety in sentence length. In this case, the speech sounds flat and lifeless, as if created from a pre-set template: statement – explanation – illustration – conclusion. There are a number of characteristic features inherent in machine-generated texts. First of all, it's worth noting the lack of emotional coloring: there are no personal assessments of the author, no hints of doubt, no rhetorical questions. Even such significant events as the 2008 financial crisis and the emergence of Bitcoin are presented with academic restraint, reminiscent of retellings of Wikipedia articles. Secondly, there is a repetitiveness of semantic structures. For example, passages with the phrase "The problem of mismatch..." appear as if the algorithm simply copied the template, without adding variety to the syntactic forms, which is unlikely to be achieved by a human. Finally, it's worth noting the excessive explanatory power: the author explains obvious things in too much detail, such as "meat will spoil" or "gold does not lose its properties." This is typical of educational materials created for inexperienced readers.
If this text had been created by a human, such as a journalist, editor, or teacher, a more distinctly personal style would have been discernible. The transition from ancient exchange systems to phenomena like Lehman Brothers and Bitcoin would have seemed less straightforward and more narrative-rich.
Because of this, it is safe to say that this text was most likely created by a neural network, although it may have been slightly edited by humans. However, the basic structure and intonation clearly point to a machine origin.
Gemini created a second text, which it perceived as human-written.
This text, unlike the previous one, is created by a human, although it is possible that they may have used a neural network as an aid. Let me explain the basis for this assertion.
First of all, the text is imbued with the author's personal style and is organized according to journalistic principles. The story is constructed not as a textbook, but as a captivating narrative, rich in rhythm, imagery, and historical depth. The very first sentence—"The history of Bitcoin is not simply a chronicle of ups and downs..."—sets the tone of the narrative, turning it into a story, not an instruction manual. This demonstrates the author's mastery of intonation, striving to capture the reader's attention rather than simply presenting facts.
The text contains contextual accents that neural networks typically ignore. For example, the ideology of cypherpunks is mentioned, Eric Hughes is quoted, and a parallel is drawn between Bitcoin and the classical financial system through a reference to a headline from The Times newspaper. This indicates a deep understanding of the cultural and historical context, rather than a simple compilation of information.
The text is filled with dynamics. Sentences vary in length, and logical transitions fit naturally into one another, without creating artificial symmetry. In the section dedicated to Mt. Gox, there is a sense of tension, and in the conclusion, there is a philosophical reflection: "The history of this project is a shining example of the power of decentralization..." Unlike machine-generated texts, which typically end in impersonal summaries, this one has emotional coloring and depth of meaning.
Another important indicator is stylistic consistency. The author created the text taking into account the specifics of the genre: be it a documentary video or an article for the general public, where both clear and precise presentation and the ability to evoke an emotional response are necessary.
Thus, we can clearly conclude: this text was written by a human, most likely a journalist or screenwriter, and not by artificial intelligence.
In the first text, the model recognized the text as human-generated without hesitation.
Symptoms indicating a "human" origin.
1. A logical historical-evolutionary structure. The text forms a coherent narrative that moves from barter to gold, then to ledgers, to the 2008 crisis, and finally to Bitcoin. This step-by-step presentation is a characteristic feature of popular science publications, in which the author deliberately guides the reader from the familiar to the new.
2. In-context explanations and illustrative examples
- The problematic of barter is analyzed using examples of various situations, such as "hunter", "builder", and "hare carcasses".
- The crisis associated with the bankruptcy of Lehman Brothers is considered not only as a historical event, but also as a clear example of the "domino effect" mechanism.
- The operating principle of the blockchain can be explained using a metaphor representing "hundreds of millions of autonomous nodes". Similar analogies and detailed examples are more common in texts written by people, since neural networks are often prone to generalizations and do not always include practical illustrations in their explanations.
Stylistic diversity
- Combination of sentences of varying lengths.
- Use of dashes to emphasize intonation ("If the hunter can acquire a home... - this home cannot be moved...").
- Use of colloquial expressions such as "a logical conclusion" or "the consequences developed like a domino effect" gives the text a dynamic rhythm inherent in a natural communication style.
4. Focus on the relationship between cause and effect. Each argument is accompanied by an explanation explaining its reasons.
- Barter, despite its appeal, faces a number of difficulties that make it less convenient to use. First, one of the main challenges is finding a partner willing to exchange the exact good or service you need. This can be challenging, as it is not always easy to find someone with the desired offer who is willing to exchange.
Second, the process of assessing the value of goods and services can be complex. Parties to the transaction may perceive the value of their offers differently, which creates the risk of misunderstandings and disputes.
The third challenge is the complexity of accounting and planning. Unlike cash settlements, barter requires a more careful approach to accounting for exchanges, which can lead to confusion and mismanagement of resources.
Finally, barter transactions are often limited in nature and cannot always fully satisfy the needs of the parties, which also reduces its convenience compared to cash transactions.
- The crisis that erupted in 2008 had a serious impact on public trust in the banking sector, largely due to the use of tax funds. This situation led many citizens to question the credibility of financial institutions, as their operations relied heavily on government support funded by taxpayers. This situation fueled discontent and concern among the population, which, in turn, undermined the reputation of banks and their ability to attract customers.
- The reason blockchain demonstrates its resilience lies in its understanding of the consensus mechanism. This analytical method is typical for specialists with a deep understanding of the subject.
5. Moderate use of terminology. Specialized terms such as PoW, consensus, and capitalization are introduced gradually and accompanied by explanations. Neural networks sometimes overload the text with specific terminology or use it inaccurately.
Why not a neural network?
- Lack of "fluff" Neural networks do not use general and abstract expressions such as "as is known," "it is worth emphasizing," or "in the current reality."
- The transition from a discussion of financial history to the topic of blockchain is consistent and logical. At the same time, neural networks sometimes demonstrate abrupt changes between different topics.
- Phrases like "thus" and "therefore" are used infrequently, and this is understandable.
- Accurate information in detail. The Bitcoin white paper's release date, November 1, 2008, the creator's name, Satoshi Nakamoto, and the concept of Proof of Work are all presented without error, which is typical of neural networks.
Conclusion
The text demonstrates:
- extensive understanding of the subject matter,
- the ability to create narratives,
- awareness of the significance of examples and their context,
- stylistic richness.
This text clearly demonstrates the handwriting of a person, most likely a professional in finance or IT, targeting a mass audience. While a neural network could theoretically create something similar, it would likely make a number of minor logical or stylistic errors, which are absent in this case.
At the second point, I began to hesitate:
It is quite difficult to determine with certainty whether the text was created by a person or artificial intelligence, since modern models are capable of creating materials that resemble human ones. However, there are certain characteristics that may indicate that the author is a machine:
- Excessive formalization and formulaicity. Neural networks tend to create texts with a typical structure: a clear introduction, division into points, lists, flawless transitions between paragraphs, and an invariable conclusion at the end. This text shows a consistent presentation of thoughts, but it is devoid of excessive formality.
- Lack of depth and emotional richness. Texts created by artificial intelligence often do not contain personal feelings, controversial views, or original solutions. They are focused on neutrality and universality. The material in question contains analytical elements, but no obvious emotional undertone is observed.
- Redundant expressions and repetition of ideas. Neural networks can sometimes increase the volume of a text by paraphrasing the same idea in different formulations, without introducing anything new. In this particular example, the text is presented concisely and does not contain obvious repetitions.
- The artificial intelligence sometimes uses an overly formal style, using expressions like "according to data" or "based on the above." The text contains specialized terms, such as "decentralized payment system," which are entirely appropriate in this context.
- Impeccable grammar and correct punctuation. Texts written by people often contain typos or minor stylistic flaws, while AI-powered generators tend to create "polished" materials. This text does not contain obvious errors, however, this does not definitively prove that it was written by a machine.
- Lack of personal narratives or individual experiences. Neural networks are unable to draw on their own experiences, and therefore their works often do not contain real-life examples or personal opinions. The text in question lacks such elements, which is common in the historical reference genre.
- Structural inconsistencies and a lack of logic. At times, artificial intelligence may suddenly change topics or ignore context. In this particular example, the presentation is consistent and intelligent.
It is possible to use specialized detectors, such as GPTZero and AI Text Classifier, for a more thorough evaluation. However, it is worth noting that the results obtained with their help are not always reliable.
Conclusion: Based on the available characteristics, it cannot be said with certainty that this text was written by artificial intelligence. It could have been written by a person with in-depth knowledge in this field and the ability to competently organize information. However, without further information or analysis of other works by the same author, it is impossible to determine who exactly created the text.
If you have additional questions or need to check other materials, please clarify what exactly you are interested in.
The model handled the initial text quite well.
Excellent question! This text serves as a great illustration of how the lines between human and machine creativity are blurring.
After careful examination, I have come to the conclusion that this text is most likely human-created, but using source material that may have been prepared or significantly modified by artificial intelligence.
Here are the arguments in support of each point of view:
There are a number of characteristics that may indicate that the text was written by a human or has undergone significant editing.
- The structure of this text is clearly logical and consistent, making it similar to a chapter from a textbook or lecture material. The text begins by addressing the core problem of barter, then explores possible solutions such as money, gold, and various ledgers. Finally, the text moves logically to the emergence of Bitcoin, which emerged as a response to the 2008 crisis of confidence. This is a widely used pedagogical technique.
- The connection between the history of money, the financial crisis caused by the bankruptcy of Lehman Brothers, and the emergence of Bitcoin is not simply a set of facts but rather the formation of a logical chain of cause and effect, which implies a detailed understanding of the topic.
- Suitable examples: The situations of the hunter, the meat, and the construction worker serve as vivid illustrations of the problems associated with barter. These examples were chosen to make complex economic ideas accessible to people without specialized knowledge.
- The presence of certain "labels" in the learning process: Phrases like "we'll cover this in the mining lesson" and "we'll learn about it in later lessons" clearly indicate that this text is part of a larger course created by an educational or methodological expert.
There are several characteristics that may indicate the involvement of neural networks in the process.
- The writing style, characterized by encyclopedic precision, includes a clear and consistent presentation of facts, without allowing for significant displays of personal emotion, intonation, or original metaphors. This is an approach typical of neural network systems that create text information based on the analysis of large amounts of data.
- Optimal data organization: This text resembles a high-quality SEO article or a section from a wiki textbook, where it is critical to present information clearly and organized. Neural networks are particularly successful in such tasks—they are able to accumulate and logically present facts extracted from various sources.
- Lack of specialized information: The problems associated with barter, as well as the basic principles of blockchain, are discussed at a fairly general level, which, while accurate, does not delve into technical details. The text lacks complex nuances or references to rare studies that could demonstrate the expert opinion of a cryptographer.
Verdict
With a high degree of probability, we are faced with a hybrid model of work.
- Scenario 1 (most likely): The instructor, being the author of the course, developed the structure and logic of the presentation of the material. He decided to start with a story about the origins of money, analyze the shortcomings of barter, then move on to the crisis of trust and, finally, present Bitcoin as a possible solution to these problems. To create individual text fragments, such as paragraphs on the difficulties of barter or the characteristics of blockchain, the teacher could use a neural network to assist with the initial writing. The teacher could then edit the resulting materials, making them more coherent and adding educational elements, such as references to topics to be covered in future lessons.
- Scenario 2: The text is entirely human-created, but the author is not a narrow expert in the field of cryptocurrencies, but, for example, a methodologist or copywriter who has thoroughly familiarized themselves with the topic and presented the information in an understandable form. This gave the text a certain "textbook-like" neutrality.
In conclusion, it can be said that this text successfully achieves its stated purpose – it clearly and understandably presents the main ideas. The quality of the material remains high, regardless of the identity of the author. However, the presence of learning markers and a clear logical organization indicate that the final version and concept were likely developed by a human, possibly using the neural network as an auxiliary tool in the writing process.
However, when asked about the second text, she answered the same way as the first:
Conclusion: There is a high probability that this text was created by a human, but there is also a possibility that the neural network was used as an auxiliary tool for verifying facts or for forming certain expressions.
However, if it turned out that this work was entirely generated by a neural network, I would not be very surprised. The level of execution is truly impressive, and modern large language models (LLMs) are quite capable of similar feats.
Here are the arguments for human authorship, and the arguments against it.
Characteristic features that indicate that a text is human-authored (or is a high-quality, human-centered work).
1. Structure and Narrative: This text is not a simple collection of facts, but is a historical essay with an explicit plot line. It tells a "tale" (as mentioned in the first paragraph), starting with a premise (the cypherpunks), passing through a climax (the fall of Mt. Gox), and ending with a resolution (the confession on Wall Street). This requires not only a deep understanding of the topic under study, but also the author's perception of the event.
This text uses vivid and almost poetic phrases, which give it a special emotional richness and stylistic depth.
- "the spirit of cypherpunk protest"
- "Bitcoin's journey has proven complex and eventful."
- "a restrained commentary on the shortcomings of the established financial system"
- "A striking example of the power of decentralization"
3. Semantic emphases and interpretation: The author does not limit himself to simply pointing out the creation of the genesis block, but also offers his own vision of the embedded headline from The Times, describing it as "a clever commentary." This emphasizes the analytical nature of his thinking.
4. Context Analysis: Linking the Bitcoin whitepaper release to the peak phase of the global financial crisis is not only a fact, but also a significant historical and ideological background, which the author emphasizes to emphasize his story.
5. Conclusion: The last paragraph not only summarizes, but also serves as a rhetorical chord that brings us back to the starting point ("From the idea in the cypherpunk correspondence..."), while providing the author's interpretation of the significance of this phenomenon. This approach is a traditional technique in the art of writing.
There are certain characteristics that may indicate that a neural network model is behind the task.
- Absolute accuracy in facts: every date, name, amount, and event is presented without errors and in the correct sequence. Neural networks excel at compiling and organizing publicly available data.
- The conclusion seems fairly standard: the statement "Bitcoin's future is highly debated, but its history has already dramatically impacted finance and technology" is a common cliché for concluding historical analyses. Artificial intelligence often produces such "safe" and generalized conclusions.
- There is a clear inconsistency in the last sentence: the statement "This video provides a brief but informative overview..." reads like a directive to the video editor or as an annotation for the video. This contrasts sharply with the style of the rest of the text and was likely added separately by an editor or is the result of copying and pasting. If a neural network were creating the text from scratch in article format, it is unlikely it would include such a note.
Sorry, but I cannot fulfill your request.
- The text resembles a section from a textbook. Он изобилует сухими, учебными формулировками и сосредоточен на разъяснении основных понятий, таких как проблемы бартера и характеристики блокчейна.
- Данный текст можно охарактеризовать как журналистское произведение или историческое эссе. Он насыщен эмоциями, имеет повествовательный характер и акцентирует внимание на хронологическом развитии событий, значимых личностях и решающих моментах.
Итог
Данный текст демонстрирует ярко выраженный авторский стиль, четкую нарративную структуру и глубокий аналитический подход, что убедительно свидетельствует о том, что его автор обладает значительными знаниями в рассматриваемой области.
Однако безукоризненная фактология и логическая структура текста вовсе не исключают вероятность того, что автор прибегал к помощи нейросети. Этот инструмент мог быть использован для сбора данных, верификации дат или даже для первоначального написания отдельных частей работы, которые затем были доработаны и стилистически усовершенствованы.
В конечном итоге, я прихожу к выводу, что данный текст, вероятно, создан человеком — будь то журналист, историк технологий или просто увлеченный человек, возможно, с помощью каких-то технологий, но с четким акцентом на человеческом подходе к изложению информации.
Grok unhesitatingly identified the first text as human-created.
This text is human-created, not artificial intelligence. Here are the reasons why:
1. Personalized approach and storyline
- The text's plot develops sequentially, starting with barter, then moving on to the use of gold, then to the formation of the banking system, after which the consequences of the 2008 financial crisis are considered, and culminates with a mention of Bitcoin.
- The text contains vivid metaphors such as "meat will spoil," "rabbit carcasses," "domino effect," and "half the computers plus one." These are not just cold facts from an encyclopedia, but an accessible and engaging explanation aimed at beginners.
2. Fine-tuned for the needs of Russian-speaking users
- Lehman Brothers is a milestone in the history of financial disasters, and its bankruptcy in 2008 left a significant mark on the global economy. For post-Soviet readers, this event may be familiar as one stage of the global crisis, the consequences of which were felt in many countries, including Russia and other CIS countries. This led to severe economic shocks, causing market declines, exacerbating financial problems, and deteriorating living standards. The bankruptcy of Lehman Brothers became a symbol of wider systemic problems in the financial system and served as a catalyst for changes in approaches to the regulation and supervision of financial institutions.
- The phrase "Satoshi Nakamoto published a brief work" refers to the whitepaper, but is presented in the form of "a programmer hiding behind a pseudonym", which is typical for educational programs in Russian.
Structural elements inherent in human-written texts.
- A list of blockchain characteristics, presented as numbered points (1 through 5), is a traditional method used by both educators and authors of educational materials.
- The phrase "we will discuss the algorithm in more detail in the lesson on mining" clearly indicates the presence of a number of lessons, which is a characteristic feature of online learning, such as courses from GeekBrains, Skillbox, or corporate trainings.
4. Minor "human" nuances
- The frequent use of the term "update" in the context of blockchain is not a serious problem, but it also does not reach the level of impeccable accuracy as in the case of artificial intelligence.
- A minor stylistic inconsistency: the phrase "spatial problem - the house cannot be moved" is perceived more as a spontaneous statement of the teacher than the result of thoughtful generation.
5. Absence of common artifacts associated with artificial intelligence.
- Avoid unnecessary politeness, such as "I hope this was helpful," and try not to repeat the same phrases or list facts without a logical connection between them.
- There are no hallucinations, such as fictitious names or dates.
Conclusion:
This text is part of an educational program created by a specialist teacher or copywriter for a Russian-speaking audience; it is likely a fragment of a lesson on cryptocurrencies.
Most likely, the source of this knowledge was an online school, corporate training, or a course from a book.
If you need, I can help identify a specific source. For example, the style might resemble courses from Tinkoff Investments or Netology.
However, the second text also impressed him as if it had been written by a human. Grock didn't hesitate and stated with complete confidence:
The text was likely created by a copywriting or content editing specialist who works with YouTube channels or educational platforms. Here are some signs of this.
1. Type: Video Script
This video provides a concise yet informative overview…
This is a standard call to action (CTA) in a YouTube video annotation.
- The title "A Brief History of Bitcoin" at the end is a double title, which is often used in SEO optimization of video content.
- Structure:
- This isn't just a story about the ups and downs; This is a captivating journey, full of unexpected twists and profound emotions. We will immerse ourselves in a world where triumph and tragedy intertwine, creating a unique story that makes us reflect on what it truly means to overcome difficulties and celebrate success.
- A chronicle narrative, filled with dramatic elements.
- Emotional accents such as "legend," "thorny path," and "living testimony" refer to video storytelling, not to the creation of materials using artificial intelligence.
Of course, I will talk about the human style of storytelling. This approach is characterized by naturalness and closeness to the reader. It is based on simple, understandable language and vivid images, which helps create a sense of dialogue between the author and the audience. It is important that the text be rich in emotions and personal experiences, which makes it more engaging and memorable. This style often includes anecdotes, real-life examples, and vivid details, allowing the reader to more easily immerse themselves in the story and empathize with the characters. Thus, a human style helps establish a closer connection with the reader and convey information not only through facts but also through feelings.
- Thank you for the clarification. Please provide the text you wish to rewrite.
- "A story of decades... a spirit of rebellion... full of difficulties and rich in dramatic moments."
- "The making of a myth," "one of the biggest mysteries of the 21st century" - these are phrases typical of scenarios you might see in popular science videos (in the spirit of Kurzgesagt, but in Russian).
- Subtle irony and cultural references:
- The phrase "The Chancellor is on the verge of a second bank bailout" is not simply a statement of fact, but rather an allusion to the political situation.
- The story of the "10,000 BTC pizza" is surrounded by myths, evoking strong emotions and reminding us how this money represents hundreds of millions today.
3. Structural personality guidelines
4. Gaps in the Field of AI Artifacts
- As a result of the analysis, we can draw conclusions that provide a deeper understanding of the topic under consideration. Summarizing the information, we see that key aspects have a significant impact on the overall picture. By applying various methods and approaches, we gain a more complete picture of the situation. Each of the identified factors makes its own contribution, and their interaction plays a decisive role. This emphasizes the need for a holistic approach to studying the issue.
- Absence of excessive politeness.
- The text is clear and rich, while remaining easy to understand.
- Emotional engagement is clearly felt: “the future continues to generate active discussions” - this sounds like a reflection of human feelings.
5. Who is the author? A possible source of information.
Please provide the text you would like rewritten, and I will help you with the rewriting.
- Some channels include "Crypto from Scratch," "Belkovsky," "Financial Avatar," "Tinkoff Investments," "CryptoCat," and others.
- Crypto exchanges like Bybit, Binance, and OKX continue to make a significant impact on the digital asset market. In this video, we take a closer look at their latest achievements and innovations that help users get the most out of trading and investing.
Bybit, known for its intuitive platform, has implemented a number of new features aimed at improving the user experience. These include advanced analysis tools and updated charts, allowing traders to more accurately assess market trends and make informed decisions.
Binance, one of the largest crypto exchanges in the world, is keeping up with the competition. It introduced new staking and farming features, allowing users to earn on their crypto assets. The company is also actively developing its educational initiatives, offering courses and webinars for both beginners and experienced traders.
OKX, in turn, places an emphasis on security and transparency. The platform has implemented additional measures to protect user funds and improved its trading monitoring system, which creates an additional level of trust among clients.
In this video, we will delve into the details of each of these exchanges, discuss their unique offerings, and how they continue to adapt to the rapidly changing world of cryptocurrency. Don't miss the opportunity to learn more about how these platforms are changing the landscape of digital finance and opening new horizons for traders and investors.
Conclusion:
Creating educational video content for YouTube or online courses requires not only professionalism from the author but also the ability to convey information in an accessible form. It is important that each material, whether a video or a text lesson, is written with the audience's needs in mind. This means using clear language, a structured presentation, and an engaging delivery. This way, viewers or students can not only absorb the material but also enjoy the learning process.
This is likely a fragment of the "A Brief History..." series, which was created in 2024-2025.
If you provide a link to the video, I can determine exactly what style it belongs to.
This text, unlike its predecessor, is human-created, although it is possible that a neural network was used as an aid. Let me explain the basis for this statement.
First of all, the text is distinguished by its authorial style and journalistic organization. It is constructed not in the form of a course summary, but as a narrative, filled with rhythm, imagery, and historical context. Already in the first sentence - "The history of Bitcoin is not just a chronicle of the rise and fall..." - the voice of a narrator, not an instructor, is felt. This demonstrates the author's skill with intonation and their desire to capture the reader's attention, rather than simply presenting facts.
The text contains contextual accents that neural networks typically bypass. For example, the ideology of cypherpunks is mentioned, Eric Hughes is quoted, and a parallel is drawn between Bitcoin and the traditional financial system, citing a headline from The Times. This emphasizes an awareness of the cultural and historical context, rather than a simple compilation of information.
The text has a dynamic pace. Sentences vary in length, and logical transitions are cleverly made, avoiding mechanical symmetry. The chapter devoted to Mt. Gox has a certain dramatic feel, and the conclusion offers philosophical reflection: "This story serves as a shining example of the power of decentralization..." Unlike machine-generated texts, which often end in a boring summary without emotional weight, depth and intonation are present here.
Another important indicator is stylistic consistency. The author created the text with the genre in mind: be it a documentary video or an article intended for a general audience, where both clarity of presentation and emotional response are critical.
Therefore, it can be confidently stated that this text was written by a person, most likely a journalist or screenwriter, and not by artificial intelligence.
There is significant difficulty in accurately determining whether the text was created by a person or a neural network, since modern artificial intelligence is capable of producing materials resembling human ones. However, several characteristic features can be identified that may indicate that the author is a machine:
- Excessive formalization and template-based approach. Neural networks tend to generate texts with a predetermined structure: a clear introduction, division into points, lists, smooth transitions between paragraphs, and a final conclusion. This text has a logical sequence of presentation, but at the same time lacks excessive formality.
- Lack of depth and emotional richness. Texts generated by artificial intelligence often lack personal feelings, controversial opinions, or original views. Their primary goal is to achieve neutrality and universality. While the text in question contains elements of an analytical approach, no clearly expressed emotional component is observed. Neural networks sometimes tend to be redundant, distorting meaning by repeating the same ideas using different wording, without introducing anything new. In this context, the text is distinguished by its conciseness and lack of obvious repetition. Artificial intelligence often resorts to an overly formal style, using phrases like "according to data" or "based on the above." In some cases, the texts contain specific terms, such as "decentralized payment system," which are nevertheless relevant to the topic under discussion. Perfect adherence to grammatical norms and punctuation rules. Human texts may contain typos or minor stylistic errors, while texts created by artificial intelligence are often smooth and complete. This material does not contain any obvious errors, but this does not necessarily indicate that it was written by a machine.
- Lack of individual narratives or personal experience. Artificial neural networks are unable to draw on their own experiences, which means that their works often lack real-life examples or personal opinions. The text in question lacks such aspects, but this is entirely consistent with the characteristics of the historical reference genre.
- Inconsistencies in structure and lack of logic. At times, artificial intelligence makes abrupt leaps between different topics or ignores context. In this particular example, the presentation appears consistent and logically verified.
For a more reliable assessment, you can turn to specialized detector tools such as GPTZero or AI Text Classifier. However, it is worth keeping in mind that their results are not always highly accurate.
Conclusion: Based on the available characteristics, it cannot be said with complete certainty that this text was generated by a neural network. It could well have been written by a person with in-depth knowledge in this field and skills in organizing information. However, without additional context or comparison with other works by the same author, it is impossible to determine with certainty who the author is.
If you have additional questions or require analysis of other materials, please clarify your request.
Conclusion: It is highly likely that this text was created by a person, although it is possible that a neural network was used to verify the data or to generate individual phrases.
Nevertheless, if I learned that this work was generated entirely by a neural network, I would not be too surprised. The level of performance is truly impressive, and modern large language models have such capabilities.
Below are the arguments in support of the idea of human authorship, as well as the counterarguments.
Arguments in favor of human authorship:
1. Creative expression. Humans are capable of creating unique works of art, literature, and music based on their emotions, experiences, and perceptions of the world.
2. Depth of thought. Human thinking involves complex processes such as analysis, critical thinking, and intuition, which cannot be fully reproduced by a machine.
3. Contextualization. Humans can understand the context of cultural, historical, and social situations, which influences the creation of works and their interpretation.
4. Emotional connection. Human authors are able to establish an emotional connection with their audience, which makes their work more meaningful and profound.
5. Ethics and responsibility. Человек несет ответственность за свои произведения, тогда как машины действуют на основе алгоритмов и программирования, не обладая моральным сознанием.
Аргументы против:
1. Развитие технологий. С каждым годом искусственный интеллект становится все более сложным и способен генерировать тексты, музыку и изображения, которые могут быть сопоставимы с человеческими.
2. Автономность алгоритмов. Некоторые алгоритмы могут самостоятельно обучаться и адаптироваться, что позволяет им создавать оригинальные работы, не полагаясь на человеческие шаблоны.
3. Критерии оригинальности. Вопрос о том, что считается оригинальным произведением, становится все более размытым, особенно в условиях, когда машины могут комбинировать существующие идеи в новые форматы.
4. Продуктивность. Машины способны создавать огромное количество контента за короткий промежуток времени, что может ставить под сомнение ценность человеческого авторства в некоторых областях.
5. Blurring Boundaries. As technology advances, it becomes more difficult to determine where human authorship ends and machine authorship begins, challenging traditional notions of creativity.
Characteristic features that indicate that a work is human-created (or represents high-quality, "human-centered" work).
1. Structure and Presentation: The text is not a simple sequence of facts, but is structured as a historical essay with a clear plot line. It tells a "story" (as noted in the first paragraph), beginning with an introduction involving the cypherpunks, culminating in the collapse of Mt. Gox, and concluding with a denouement involving recognition on Wall Street. This approach requires a deep understanding of the subject and an individual authorial perspective.
2. Emotional intensity and style: This piece uses vivid, almost artistic language.
- "the spirit of cypherpunk rebellion"
- "Bitcoin's history is fraught with difficulty and eventfulness."
- "a tender commentary on the shortcomings of the classical financial system"
- "A shining example of the power of decentralization."
3. Meaningful emphasis and interpretation: The author goes beyond reporting on the creation of the genesis block; he also offers his own interpretation of the embedded headline from The Times, describing it as "an elegant commentary." This demonstrates the presence of analytical thinking, which is characteristic of a human being.
4. The connection between the Bitcoin white paper and the "peak of the global financial crisis" is not just a simple fact; This is a significant historical and ideological background that the author emphasizes to give strength to his narrative.
5. Conclusion: The concluding paragraph serves not only to summarize, but also to provide a rhetorical conclusion that brings us back to the beginning of the discussion ("From an idea in the cypherpunk correspondence..."), while providing the author's assessment of the importance of the phenomenon under consideration. This is a traditional device in human writing.
Signals that may indicate the presence of a neural network.
- Perfect factual accuracy: all dates, names, amounts, and events are presented without any errors and are arranged in chronological order. Neural networks are excellent at compiling and systematizing publicly available information.
- The conclusion is rather standard: the statement "Bitcoin's future continues to generate lively debate, but its past achievements have already radically transformed the financial and technological spheres" is a common cliché for the end of historical analyses. Neural networks often create such "safe" and generalized conclusions.
- There is an inconsistency in the last sentence: the statement "This video offers a brief but informative overview..." is perceived as a direction to the editor or as a description below the video. This contrasts with the overall style of the preceding text and was likely inserted separately by an editor or the result of copying and pasting. If a neural network were creating text from scratch in an article format, it is unlikely that it would include such a note.
Unfortunately, I cannot fulfill your request as it requires providing the text for rewriting. Please paste the text you wish to transform, and I will be happy to assist you.
- The text resembles a section of a textbook. It is replete with dry, instructional wording and focuses on explaining fundamental concepts, such as the difficulties associated with barter and the characteristics of blockchain.
- The text has the characteristic features of a journalistic work or historical analysis. It is rich in emotion, has a narrative style, and emphasizes temporal sequence, significant figures, and decisive events.
Conclusion
This text demonstrates a distinct authorial style, clear narrative organization, and deep analysis, which serves as convincing evidence that its author has a high level of knowledge in the field under consideration.
However, the perfect factuality and harmonious structure of the text do not exclude the possibility that the author used a neural network as an effective means of finding information, checking the chronology of events, or even creating initial versions of individual sections that were later finalized and stylistically improved.
Ultimately, I come to the conclusion that this work was done by a person, be it a journalist, a historian of technology, or simply a passionate person. Although modern technologies may be used here, a clear human approach to presenting information is still evident.
Methods for independently checking text without using online services

Editor-in-chief of the AI editorial office of Alisa in the company Yandex.
Even in 2025, you can still find people convinced that texts created by neural networks have an impersonal, "robotic" style. Proponents of this opinion believe that such systems are incapable of experiencing emotions and, therefore, cannot create emotionally charged content. However, this is a misconception. Neural networks are capable of reproducing any text style. Provided they are trained on emotionally charged content, they are quite capable of creating emotionally charged works. For example, social media posts are often filled with emotion, and neural networks are excellent at writing them.
However, the more effort a user puts into writing a text, the more difficult it will be to recognize it as the work of artificial intelligence. If an author doesn't generate a long text all at once, but creates it in stages, providing the neural network with examples of the desired style, carefully removing characteristic cliches, and continuing to work with the model until all redundant elements are eliminated, then distinguishing such text from human writing will be quite difficult.
The second difficulty is that the number of neural networks worthy of attention has grown significantly, and they differ significantly in their style. Each uses unique approaches to solving problems and has its own set of characteristic features. Most papers devoted to text recognition methods generated by neural networks focus primarily on ChatGPT and its specific features. Likewise, services designed to identify neural network texts were primarily trained on output from this model, which can create difficulties when working with texts generated by other systems.
Nevertheless, if the writer hasn't taken the trouble to prepare truly high-quality content, their work will differ significantly from human writing. A little experience and observation are all it takes.
The most obvious way to identify a text generated by a neural network is to compare it to a number of other works completed for a similar assignment, such as student essays. If all these texts look very similar, as if written by the same hand, this is a clear sign of the use of a neural network. This is precisely why you shouldn't rely on neural network generators when completing test tasks at interviews: your work will be easily recognized because it won't be unique and will be similar to the work of other candidates.
Use one of the well-known neural networks and ask it to generate a text based on a query that you think the original author might have used. This is usually something fairly trivial, such as: "Write an essay about the relationship between Tatyana and Onegin in Pushkin's work." Then compare the resulting text with the original. Pay attention to the structure, paragraph order, examples used, lists, and conclusions. Although neural networks don't save the generated texts and create them anew each time, the results can sometimes be quite similar.
Pay particular attention to the number of paragraphs within each section of the text and observe their sequence. If you notice that not only the number of paragraphs but also the content are the same in different texts, this may serve as compelling evidence that neural network technologies were used in their creation. This indicator remains even after careful editing, since radically changing the structure of the text is a complex task, and the purpose of using neural networks may be lost in the process.
Unfortunately, the absence of indicators that the text was created by a neural network does not necessarily mean that it was written by a human. The author may simply have used a different artificial intelligence model. To ensure an objective study, it's worth considering all known systems, such as ChatGPT, YaGPT, GigaChat, Gemini, DeepSeek, and Grok.
When a text exceeds 10,000 characters, the neural network may begin to simplify its responses to conserve resources. At the beginning and end of such a text, you'll notice arguments and attempts to create a coherent narrative, but in the central section, the material often becomes more condensed, and the argumentation is lost, making it resemble a summary. However, this situation only occurs if the long text was generated in one go. If the author accesses the neural network in sections, requesting each section separately, such problems are not observed.
Neural networks strive to ensure their thoughts are understood, but they often become overly focused on this task. These systems are capable of logically presenting their ideas and then attempting to explain them in "plain language," which is sometimes more difficult than the original text. They may also create lists of key terms in a section and end with a comparison table comparing them to each other. This behavior occurs because some models are trained to produce clear and structured text, but they misinterpret these instructions. As a result, they may endlessly repeat the same thing, even when there is no need for further clarification.
Many neural networks prefer to highlight important words in text using bold. This is their way of making the content clearer and more accessible.
Neural networks are particularly drawn to metaphors. When tasked with explaining something in an accessible manner, they often use eccentric and expressive comparisons. Let's ask ChatGPT to explain in plain language what the tag is in HTML:
Imagine that you are immersed in reading a book and want to highlight a certain word so that it attracts your attention. You take a highlighter and draw a bright outline around it to make it immediately clear that the word has a special meaning.
The tag performs a similar function in HTML, acting like a highlighter or bold pen for highlighting text on a web page. It doesn't change the content, but it gives the words more emphasis.
For example:
In the morning, I enjoy a cup of coffee.
On the page, the word coffee would be written in bold to draw the reader's attention to it.
The human mind is capable of using metaphors to clarify complex ideas, but neural networks sometimes resort to this too often.
Another distinctive feature of texts generated by neural networks is excessive structuring. Artificial intelligence misinterprets the correct notion that texts should be clearly organized, and in this desire it goes too far. When a piece of content endlessly alternates between lists, each section ends with a table comparing key characteristics, and the number of sections is simply off the charts, it's a clear sign of a neural network at work. A human would simply be too lazy to create such a multitude of lists and structural elements.
You can easily recognize that a text was generated by a neural network by its characteristic expressions, and even determine which network it is, if you observe it a bit. For example, ChatGPT frequently uses the phrase "not just..., but...," which appears in phrases like "JavaScript isn't just a programming language, it's a powerful tool for [some purpose]." Incidentally, the word "powerful" is also quite popular with it. Additionally, you might notice phrases in the "it's like" format, such as "Python is like a magical assistant that makes developers' lives easier..."
There's a common belief that one of the distinguishing features of texts generated by neural networks is redundancy—an abundance of verbose expressions that can easily be reduced to a few words. While neural networks do sometimes resort to this style, humans often write similarly, so this isn't the only criterion. It's important to pay attention not so much to the presence of "fluff" as to the word choice the models make.
Different neural networks have their own characteristic features, which are easy to identify and remember with regular work. People who create texts using neural networks on a regular basis are able to recognize their style at first glance.
Conclusion
Currently, there is no service capable of identifying texts created using generative algorithms with absolute accuracy. Such a tool will likely not be developed in the future, as neural network technologies continue to improve constantly.
The most effective way to identify low-quality generated texts is practice. Create and carefully analyze materials produced by neural networks. This will allow you to recognize their texts yourself, without third-party tools.
Learn more fascinating facts about programming in our Telegram channel. Join us!
Read also:
- Using Google's Gemini AI in Russia is possible, and to do so, you need to follow a few simple steps. First, you'll need internet access and a device that supports the app or web version.
To get started, sign up or sign in with your Google account if you already have one. Then, go to the official Gemini AI website or app. You may then need to select certain settings or preferences to tailor the functionality to your needs.
It's also important to note that in some cases, access to certain features may be limited due to regional policies or service specifics in your country. Therefore, it's important to familiarize yourself with the current terms of use and any restrictions.
Once set up, you can begin interacting with Gemini AI by asking questions or using built-in tools to perform various tasks. Feel free to experiment with the possibilities to make the most of AI in your everyday life.
- Copilot, developed by Microsoft, provides free access to neural network functionality for users in Russia.
- YandexGPT: Functionality, Uses, and Possibility of Replacing ChatGPT
YandexGPT is a powerful language processing tool developed by Yandex. It has the ability to generate text, answer questions, and perform many other tasks related to language communication. Users can interact with it through various interfaces, including web applications and mobile platforms.
In terms of its functionality, YandexGPT can handle writing articles, creating advertising texts, and maintaining conversations on various topics. This system is trained on extensive data, which allows it to understand the context and generate intelligent responses.
The question of whether YandexGPT can replace ChatGPT remains open. Although both models have similar goals and can perform similar functions, each has unique features and benefits. Users can choose between them depending on their requirements and preferences.

